a new humanized mouse model of epstein-barr virus infection … · 2017-04-13 · a new humanized...
TRANSCRIPT
A New Humanized Mouse Model of Epstein-BarrVirus Infection That Reproduces Persistent Infection,Lymphoproliferative Disorder, and Cell-Mediatedand Humoral Immune Responses
Misako Yajima,1,a Ken-Ichi Imadome,1,a Atsuko Nakagawa,2 Satoru Watanabe,3 Kazuo Terashima,4
Hiroyuki Nakamura,1 Mamoru Ito,6 Norio Shimizu,3 Mitsuo Honda,5 Naoki Yamamoto,4,5 and Shigeyoshi Fujiwara1
1Department of Infectious Diseases, National Research Institute for Child Health and Development, 2Pathology Laboratory, Department of ClinicalLaboratory Medicine, National Center for Child Health and Development, 3Department of Virology, Division of Medical Science, Medical ResearchInstitute, and 4Department of Molecular Virology, Graduate School of Medicine, Tokyo Medical and Dental University, and 5AIDS Research Center,National Institute of Infectious Diseases, Tokyo, and 6Central Institute for Experimental Animals, Kawasaki, Japan
The functional human immune system, including T, B, and natural killer lymphocytes, is reconstituted in NOD/Shi-scid/IL-2R�null (NOG) mice that receive hematopoietic stem cell transplants. Here, we show that these human-ized mice can recapitulate key aspects of Epstein-Barr virus (EBV) infection in humans. Inoculation with�1 � 103 TD50 (50% transforming dose) of EBV caused B cell lymphoproliferative disorder, with histopatholog-ical findings and latent EBV gene expression remarkably similar to that in immunocompromised patients. Inoc-ulation with a low dose of virus (�1 � 101 TD50), in contrast, resulted in apparently asymptomatic persistentinfection. Levels of activated CD8� T cells increased dramatically in the peripheral blood of infected mice, andenzyme-linked immunospot assay and flow cytometry demonstrated an EBV-specific T cell response. Immuno-globulin M antibody specific to the EBV-encoded protein BFRF3 was detected in serum from infected mice. TheNOG mouse is the most comprehensive small-animal model of EBV infection described to date and should facil-itate studies of the pathogenesis, prevention, and treatment of EBV infection.
Epstein-Barr virus (EBV) is a tumor virus associated
with a variety of malignancies, including Burkitt lym-
phoma, nasopharyngeal carcinoma, and Hodgkin
lymphoma [1]. It is also an etiological agent of infec-
tious mononucleosis (IM), which is characterized by
transient proliferation of EBV-infected B lympho-
blastoid cells and an excessive anti-EBV T cell re-
sponse. EBV has a unique ability to growth transform
human B lymphocytes in vitro and establish lympho-
blastoid cell lines (LCLs) [2]. EBV-transformed lym-
phoblasts express 6 nuclear proteins (Epstein-Barr
nuclear antigen [EBNA] 1, 2, 3A, 3B, 3C, and LP) and
3 membrane proteins (latent membrane protein
[LMP] 1, 2A, and 2B), and this pattern of EBV gene
expression is termed latency III. In contrast, Burkitt
lymphoma cells express only EBNA1 consistently (la-
tency I), whereas Hodgkin lymphoma and nasopha-
ryngeal carcinoma cells express EBNA1, LMP1, and
LMP2 (latency II). In vivo, EBV-transformed cells are
effectively removed by virus-specific cytotoxic T cells,
and EBV infection in immunocompetent humans is
usually subclinical, except for IM caused by primary
infection during adolescence or adulthood. However,
in immunocompromised hosts, such as patients with
AIDS and transplant recipients, EBV-infected B lym-
phoblasts can proliferate and cause lymphoprolifera-
tive disorder.
Received 28 December 2007; accepted 19 March 2008; electronically published15 July 2008.
Potential conflicts of interest: none reported.Financial support: Ministry of Health, Labour, and Welfare of Japan (grants
H18-Shinko-013 and H19-AIDS-003).a M.Y. and K.-I.I. contributed equally to this study.Reprints or correspondence: Dr. Shigeyoshi Fujiwara, Dept. of Infectious Dis-
eases, National Research Institute for Child Health and Development, 2-10-1Okura, Setagaya-ku, Tokyo 157-8535, Japan ([email protected]); or, Dr. NorioShimizu, Dept. of Virology, Div. of Medical Science, Medical Research Institute,Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8519,Japan ([email protected]); or, Dr. Naoki Yamamoto, AIDS Research Center, Na-tional Institute of Infectious Diseases, 1-23-1 Toyama, Shinjuku-ku, Tokyo 162-8640, Japan ([email protected]).
The Journal of Infectious Diseases 2008; 198:673– 82© 2008 by the Infectious Diseases Society of America. All rights reserved.0022-1899/2008/19805-0008$15.00DOI: 10.1086/590502
M A J O R A R T I C L E
New Mouse Model of EBV Infection ● JID 2008:198 (1 September) ● 673
EBV infects only humans in nature and limited animal species
under experimental conditions. It can infect cotton-top tama-
rins and induce lymphomas, which have been used as a model of
EBV-associated lymphomas [3, 4]. Nonhuman primates possess
their own lymphocryptoviruses related to EBV, and research on
the use of these virus-host systems as models of EBV infection is
currently in progress [5, 6]. Small-animal models of EBV have
also been developed, which are particularly useful when a large
number of animals are necessary. Scid mice that receive intra-
peritoneal transplants of EBV-transformed LCLs or peripheral
blood mononuclear cells (PBMCs) isolated from EBV-infected
persons develop lymphomas, which have been used as a model
of human lymphoproliferative disorder [7–9]. Recently, NOD/
scid mice transplanted with human hematopoietic stem cells
(HSCs) and reconstituted mainly with B lymphocytes were in-
fected with EBV, and the development of lymphoproliferative
disorder was described [10]. The immune response to EBV was
not studied in these scid or NOD/scid mouse models. Very re-
cently, a functional human immune system could be reconsti-
tuted in highly immunodeficient mouse strains, and these so-
called humanized mice were shown able to mount an EBV-
specific T cell response [11, 12]. These studies were, however,
performed mainly using immunological standpoints and did
not provide detailed virological data.
NOD/Shi-scid/IL-2R�null (referred to here as NOG) is a highly
immunodeficient mouse strain that was developed very recently
and that, after transplantation with cord blood HSCs, is able to
reconstitute most major components of the hematolymphoid
system, including T cells, B cells, NK cells, macrophages, and
dendritic cells [13–15]. Human T cells that develop in NOG
mice are functional in that they can be activated to display cyto-
toxic activity [15, 16]. These properties made NOG mice an ex-
cellent model of human virus infections targeting the immune
system, such as those with human T-lymphotropic virus–1 and
HIV-1 [17–20]. Here, we provide evidence that humanized
NOG mice can reproduce various key aspects of human EBV
infection and propose that they may be a valuable tool for studies
of EBV infection.
METHODS
Preparation of humanized mice. NOG mice were obtained
from the Central Institute for Experimental Animals (Kawasaki,
Japan). Protocols for experiments with NOG mice were ap-
proved by the Institutional Animal Care and Use Committee of
the National Institute of Infectious Diseases (NIID). Cord blood
was supplied by the Tokyo Cord Blood Bank after obtaining
informed consent. The use of human materials in this research
was approved by the institutional review boards of the National
Research Institute for Child Health and Development, the NIID,
the Tokyo Medical and Dental University, and the Tokyo Cord
Blood Bank. The isolation of human CD34� HSCs from cord
blood by means of the MACS Direct CD34 Progenitor Cell Iso-
lation Kit (Miltenyi Biotec), their intravenous injection
(1 � 104 to 1.2 � 105 cells/mouse) into 6 –10-week-old female
NOG mice, and the characterization of the reconstitution of the
human hematoimmune system were done as described else-
where [18, 20]. NOG mice in which the human hematoimmune
system was reconstituted are referred here as humanized NOG
(hNOG) mice.
Experimental EBV infection, quantification of viral DNA,
and detection of viral mRNAs. Virus production by EBV-
infected Akata cells was stimulated by brief treatment with anti-
IgG antibody (Dako), and culture fluid was used as inoculum
after filtration through a 0.45-�m membrane filter [21]. For vi-
rus titration, cord blood lymphocytes were plated at the density
Table 1. Primers for reverse-transcription poly-merase chain reaction to detect Epstein-Barr virus(EBV) transcripts.
Transcript, primer Sequence (5'33')
EBNA15' gatgagcgtttgggagagctgattctgca3' tcctcgtccatggttatcac
EBNA25' agaggaggtggtaagcggttc3' tgacgggtttccaagactatcc
LMP15' ctctccttctcctcctcttg3' caggagggtgatcatcagta
LMP2A5' atgactcatctcaacacata3' catgttaggcaaattgcaaa
LMP2B5' cagtgtaatctgcacaaaga3' catgttaggcaaattgcaaa
EBER15' agcacctacgctgccctaga3' aaaacatgcggaccaccagc
BZLF1 (first)5' attgcaccttgccgccaccttg3' cgcattttctggaagccacccga
BZLF1 (second)5' gaccaagctaccagagtctat3' cagaatcgcattcctccagcga
BMRF15' ctagccgtcctgtccaagtgc3' agccaaacagctccttgccca
BLLF15' gtcagtacaccatccagagcc3' ttggtagacagccttcgtatg
GAPDH5' gcctcctgcaccaccaactg3' cgacgcctgcttcaccaccctct
NOTE. EBNA, Epstein-Barr nuclear antigen; EBER, EBV-encoded small RNA; LMP, latent membrane protein.
674 ● JID 2008:198 (1 September) ● Yajima et al.
of 2 � 105 cells per well in 6-well plates and then inoculated with
serial 10-fold dilutions of virus preparation. The number of wells
with proliferating lymphocytes was counted 6 weeks after infec-
tion, and the titer of the virus in 50% transforming dose (TD50)
was determined by the Reed-Muench method [22]. EBV was
inoculated intravenously through the tail vein. EBV DNA was
quantified by a real-time quantitative polymerase chain reaction
(PCR) assay based on the TaqMan system (Applied Biosystems),
as described elsewhere [23]. Analysis of EBV gene expression by
reverse-transcription PCR (RT-PCR) was done as described
elsewhere, using the primers listed in table 1 [24].
Histopathology, in situ hybridization (ISH), and immuno-
histochemistry. Tissue samples were fixed in 10% buffered
formalin, embedded in paraffin, and stained with hematoxylin-
eosin. For phenotypic analysis of proliferating lymphocytes, im-
munostaining for CD3 (Nichirei), CD4 (Novocastra), CD8
(Nichirei), CD45RO, CD20, CD79a, CD30, Mum1 (Dako),
CD23, CD10, CD56 (Novocastra), granzyme B (Dako), and T
cell intracellular antigen 1 (Beckman Coulter) was performed on
paraffin sections. EBV was detected by immunostaining for
LMP1 and EBNA2 (Dako) and by ISH with EBV-encoded small
RNA (EBER) probe. Immunohistochemistry and ISH were per-
formed on an automated stainer (Benchmark XT; Ventana
Medical Systems), in accordance with the manufacturer’s rec-
ommendations. To determine the cell lineage of EBV-infected
cells, paraffin sections were applied to double staining with
EBER ISH and immunohistochemistry.
Detection of EBV-specific T cell response. Enzyme-linked
immunospot (ELISPOT) assay was performed with the Im-
munocyto IFN-� ELISPOT Kit (MBL), in accordance with
the instructions supplied by the manufacturer. Briefly, CD8�
T cells were isolated from PBMCs from EBV-infected hNOG
mice with the IMag anti-human CD8 Particles–DM (BD Bio-
sciences). Mixture of these CD8� T cells and an autologous
LCL were incubated with interleukin (IL)–2 in microplates
coated with antibody to interferon (IFN)–� for 17 h. Cap-
tured IFN-� was detected by use of biotinated antibody to
IFN-� and alkaline phosphatase– conjugated streptavidin
and was visualized by reaction with the BCIP/NBT chromo-
gen substrate. The unpaired Student’s t test was used for sta-
tistical analysis. IFN-� secretion in response to EBV was also
examined by flow cytometry, as described elsewhere [25].
Briefly, aliquots of murine splenocytes and an LCL were
mixed in 6-well plates in the presence of brefeldin A (10 �g/
mL) and incubated at 37°C in 5% CO2 for 17 h. After incuba-
tion, the cell suspensions were stained with phycoerythrin-
conjugated anti-human CD69, phycoerythrin–Texas red–
conjugated anti-human CD45, and phycoerythrin– cyanin
5– conjugated anti-human CD8 for 30 min at 4°C and were
fixed with 2% paraformaldehyde. Cells were then permeabil-
ized and stained with BD Perm/Wash buffer (BD Biosciences)
containing fluorescein isothiocyanate–conjugated anti-human
IFN-� for 30 min at 4°C. Stained cells were analyzed using an
EpicsXL flow cytometer (Beckman Coulter).
Figure 1. Peripheral blood Epstein-Barr virus (EBV) DNA load and body weight in humanized NOG (hNOG) mice infected with EBV. A, Infection ata high dose of virus. Six mice were inoculated intravenously with 1 � 103 TD50 of EBV. Peripheral blood EBV DNA load (upper panels) and body weight(lower panels) were then determined weekly. Each symbol in the graphs represents an individual mouse. Interruption of records indicates the death orkilling of a mouse. B, Infection at lower doses. Peripheral blood EBV DNA load (upper panel) and body weight (lower panel) of 2 mice inoculated withlow doses of EBV (black circle, 1 � 10-1 TD50; white circle, 1 � 101 TD50) are shown.
New Mouse Model of EBV Infection ● JID 2008:198 (1 September) ● 675
Figure 2. Pathological and virological analyses of Epstein-Barr virus (EBV)–infected humanized NOG (hNOG) mice. A, Photograph of an EBV-infected mouseshowing tumors in the cervical area. B, Photographs of spleens, liver, lymph node, and kidney from EBV-infected mice with lymphoproliferative disorder. Theupper left panel shows the spleen from an uninfected mouse. C, Photomicrographs of hematoxylin-eosin–stained tissues from mice with lymphoproliferativedisorder. The arrow indicates a Reed-Sternberg–like cell, and the arrowheads indicate Hodgkin-like cells. Original magnifications, �1000 for spleen, �400 forlymph node, and �200 for liver, lung, kidney, and adrenal gland. D, Immunohistochemical staining for lymphocyte surface markers (CD3, CD20, CD23, andMum1) and EBV-encoded proteins (latent membrane protein [LMP] 1 and Epstein-Barr nuclear antigen [EBNA] 2), as well as in situ hybridization for EBV-encodedsmall RNA (EBER), in a lymph node from a mouse with lymphoproliferative disorder. The bottom right panel represents double staining for EBER and CD20.Original magnifications, �200 for all except EBER/CD20, which is �400. E and F, Reverse-transcription polymerase chain reaction detection of latent-cycle (E)and lytic-cycle (F) EBV gene expression in tumors from EBV-infected hNOG mice. Spleen tumors from 3 different mice were examined for the expression ofEBNA1, EBNA2, LMP1, LMP2A, LMP2B, EBER1, BZLF1, BMRF1, and BLLF1. RNA samples from a lymphoblastoid cell line (LCL) (E) and anti-IgG–treated Akatacells (F) were used as positive controls, and an RNA sample from EBV-negative Akata cells (E and F ) was used as a negative control. Assays were done with(�) or without (�) reverse transcriptase (RT) in panel F. Expression of GAPDH was examined as a reference. G, Double staining of EBER and CD20 in the liverof an hNOG mouse that was persistently infected with EBV without developing lymphoproliferative disorder. EBER is stained navy in the nucleus, and CD20is stained brown in the membrane. Original magnification, �1000.
Detection of antibodies specific to EBV. IgM antibody to
the EBV BFRF3 protein was detected by immunoblotting essen-
tially as described elsewhere [24], except that horseradish per-
oxidase– conjugated antibody specific to human IgM (Beckman
Coulter) was used as secondary antibody. To prepare the gluta-
thione s-transferase (GST)–BFRF3 fusion protein, a DNA frag-
ment spanning the entire coding region of BFRF3 was amplified
by PCR (sense primer, 5'-GGCTCGAATTCATGGCACGCCG-
GCTGCCC-3'; antisense primer, 5'-GGCTCGGATCCATAC-
ACCATGTTTCGTGCCC-3') and inserted to the GST fusion
vector pSGENT2, to yield the plasmid pSGENT2-BFRF3. Esch-
erichia coli cells harboring pSGENT2-BFRF3 were stimulated
with isopropyl �-D-1-thiogalactopyranoside to induce the ex-
pression of GST-BFRF3, which was subsequently purified by use
of the Bulk GST Purification Module (GE Healthcare).
RESULTS
EBV infection in hNOG mice. Transplantation of human
CD34� HSCs in NOG mice and reconstitution of the human
hematoimmune system were done as described elsewhere [18,
20]. In the initial attempts at infection, 1 � 103 TD50 of the
Akata strain of EBV was inoculated into 6 hNOG mice, and EBV
DNA was demonstrated in the peripheral blood of all of them
(figure 1A). EBV DNA was first evident at 3– 4 weeks after inoc-
ulation and reached peak levels of �1 � 106 EBV DNA cop-
ies/�g of DNA. All 6 mice became seriously ill between 5 and 10
weeks after inoculation, with signs of weight loss (figure 1A),
general inactivity, and piloerection. In contrast, EBV DNA was
not detected in the peripheral blood, bone marrow, thymus,
spleen, lymph nodes, liver, kidneys, and lungs of 3 control NOG
mice that were not transplanted with HSCs but were inoculated
with the virus (data not shown). Similarly, no signs of EBV in-
fection were observed in 3 control hNOG mice that were not
inoculated with the virus (data not shown). In total, 43 NOG
mice that had been humanized with HSCs from 9 different cord
blood samples were inoculated with 1 � 103 TD50 of EBV, and
in 38 of them the results were similar to those observed in the
initial 6 mice, with high blood EBV load and severe deterioration
in their general condition. Ten of them died and could not be
examined further. The remaining 28 mice were killed, and signs
of lymphoproliferative disorder were found at autopsy (see the
below). These results demonstrate that hNOG mice can be in-
fected with EBV, with a mostly fatal outcome at this virus dose.
EBV-induced lymphoproliferative disorder in hNOG mice.
Autopsy of killed mice showed signs of lymphoproliferative dis-
order typically represented by an overt tumor in the spleen (fig-
ure 2B). In �70% (20/28) of the mice autopsied, macroscopical
signs of disseminated disease were found in the liver, lymph
nodes, or kidneys (figure 2A and 2B). Seventeen mice were ex-
amined pathologically, and 15 of them showed typical histology
of diffuse large B cell lymphoma, with remarkable similarity to
the human lymphoproliferative disorder in the immunocom-
promised hosts (figure 2C). The tissues contained occasional
immunoblasts, Reed-Sternberg–like cells, and Hodgkin-like
cells (figure 2C). Marked infiltration of large transformed lym-
phoid cells was also demonstrated in liver, lymph nodes, kid-
neys, adrenal glands, and lungs (figure 2C). Real-time PCR de-
tected high levels (�1 � 105 to �1 � 106 EBV DNA copies/�g
of DNA) of EBV DNA in these organs, and the large transformed
lymphoid cells were universally EBV positive by EBER ISH (fig-
ure 2D). Immunohistochemical analysis showed that the large
transformed lymphoid cells were of the activated B cell pheno-
type, being reactive for CD20 and CD23 and not reactive for
CD3 and CD10 (figure 2D and data not shown). They were also
positive for Mum-1, a late- and postgerminal center cell marker.
The EBER-positive cells were CD20-positive B cells (figure 2D),
and no EBER-positive T cells were identified. Immunostaining
revealed that most proliferating cells expressed EBNA2, whereas
LMP1 was expressed in only a fraction of them (figure 2D). RT-
PCR analysis of typical spleen tumors obtained from 3 different
mice showed the expression of EBNA1, EBNA2, LMP1, LMP2A,
LMP2B; and EBER, consistent with the latency III program of
EBV gene expression (figure 2E). In addition, transcripts from
lytic-cycle EBV genes, including BZLF1 (immediate-early),
BMRF1 (early), and BLLF1 (late, encoding gp350/220), were
identified (figure 2F).
Virus dose– dependent outcome of EBV infection in hNOG
mice. To examine the influence of virus dose on the outcome
of EBV infection, we inoculated serial dilutions of EBV prepara-
tion into 2 lots of hNOG mice, each consisting of 5 mice that had
been humanized with the same HSC preparation. Consistent
with the results described above, the 4 mice (2 from each lot) that
received the higher doses (1 � 103 and 1 � 102 TD50) of the
Table 2. Quantification of Epstein-Barr virus(EBV) DNA in persistently infected humanized NOGmice.
Organ
Mouse
N35-1a N35-3b
Bone marrow ND 4.1 � 104
Spleen 6.2 � 102 5.7 � 103
Liver ND 2.7 � 104
Lymph node (neck) 1.6 � 103 6.9 � 103
Lymph node (axilla) ND 2.6 � 102
Lymph node (mesentery) ND 4.1 � 102
Lungs 2.7 � 103 1.0 � 104
Kidneys 1.2 � 103 4.8 � 104
Adrenal gland 4.4 � 101 8.0 � 105
NOTE. Data are the amounts of EBV DNA measured 22weeks after infection, in copies per microgram of DNA. ND,not detectable.
a Infected at 1 � 10-1 TD50.b Infected at 1 � 101 TD50.
New Mouse Model of EBV Infection ● JID 2008:198 (1 September) ● 677
Figure 3. Surface marker expression by peripheral blood T cells in Epstein-Barr virus (EBV)–infected humanized NOG (hNOG) mice. A, Changes inthe percentages of CD3� T cells and CD19� B cells among human CD45� leukocytes (upper panels) and in the percentages of CD8� cells and CD4�
cells among CD3� cells (lower panels) after infection with EBV. Results obtained from 3 EBV-infected mice and an uninfected mice are shown. Whitesymbols indicate the percentage of CD19� cells (upper panels) or CD4� cells (lower panels); black symbols indicate the percentage of CD3� cells (upperpanels) or CD8� cells (lower panels). A vertical arrow in the graph area shows the time point at which EBV DNA was first detected in peripheral blood.B, Changes in the no. of CD8� T cells in the peripheral blood of EBV-infected hNOG mice. White symbols indicate uninfected mice, and black symbolsindicate infected mice. Note that cell no. is plotted in a logarithmic scale. C, Viral dose– dependent T cell responses in hNOG mice inoculated withserially diluted EBV. Ten-fold serial dilutions of an EBV sample starting from 1 � 103 TD50 per inoculate were injected intravenously into NOG micethat had undergone transplantation with the same lot of human hematopoietic stem cells (HSCs). Changes in the percentages of CD3� T cells amonghuman CD45� leukocytes (upper panel) and in the percentages of CD8� cells among CD3� cells (lower panel) after inoculation with EBV are shown.The viral dose for each mouse is shown in the key. D, Comparison of surface marker expression between EBV-infected mice and control mice. Twomice that underwent transplantation with the same lot of human HSCs were either inoculated with EBV or left uninfected; 10 weeks after inoculation,mononuclear cells obtained from peripheral blood or spleen were gated for the expression of human CD3 and then examined for the expression of CD8and CD4 (top panels), CD45RO and CD45RA (second from top), HLA-DR and CD4 (second from bottom), and HLA-DR and CD8 (bottom).
virus died of lymphoproliferative disorder �5–10 weeks after
inoculation. The remaining mice in both of the lots that received
lower doses (1 � 101, 1 � 100, and 1 � 10-1 TD50) survived
acute infection and appeared normal throughout the observa-
tion period of 22 weeks. Although EBV DNA was detected at
variable levels in their peripheral blood several weeks after inoc-
ulation, it returned to undetectable levels thereafter (figure 1B),
suggesting that a certain protection mechanism worked to con-
trol EBV infection. Importantly, EBV DNA could be still de-
tected in various organs, including spleen, liver, lungs, kidneys,
and adrenal glands, at the end of the observation period (22
weeks), indicating that EBV persisted in these mice (table 2).
Double staining for EBER and CD20 showed that EBV persisted
in B cells (figure 2G). Macroscopical examination by autopsy at
the end of the observation period did not reveal abnormality in
these mice, except for moderate splenomegaly found in a mouse
that received 1 � 101 TD50. These results indicate that the out-
come of EBV infection in hNOG mice varies with the virus dose;
high doses of virus tend to cause fatal lymphoproliferative dis-
order, whereas lower doses induce apparently asymptomatic
persistent infection.
EBV-specific T cell response in hNOG mice. Flow cytom-
etry analysis demonstrated a dramatic increase in the percentage
of CD3� T cells among the human CD45� leukocytes after in-
fection with EBV. This increase in T cells was accompanied by an
increase in the percentage of CD8� cells among human CD3� T
cells. These changes were seen in virtually all infected mice, and
the results from 3 mice are shown in figure 3A. The slow increase
in the percentage of CD3� cells in the uninfected mouse repre-
sents the process of humanization (i.e., the development of hu-
Figure 4. Epstein-Barr virus (EBV)–specific T cell response in humanized NOG (hNOG) mice. A, Enzyme-linked immunospot assay for the detectionof human T cells producing interferon (IFN)–� after stimulation with an EBV-positive lymphoblastoid cell line (LCL). CD8� cells isolated from EBV-infectedhNOG mice were cocultured with an autologous LCL, and IFN-�–secreting cells were counted (3, 5, 6, 7, and 8). To analyze restriction by majorhistocompatibility complex (MHC), antibody to HLA class I (anti– human HLA-ABC clone W6/32; eBioscience) (5), antibodies to both HLA class I and classII (6), antibody to HLA class II (anti– human HLA-DP,DQ,DR clone CR3/43; Dako) (7), or isotype-matched control antibody (8) were added to the culture.Control experiments included coculture of CD8� cells with an MHC-mismatched LCL (1), culture of the autologous LCL only (2), and culture of CD8�
cells only (4). Results from 2 infected mice are shown. Five hundred CD8� cells per well were cultured in the experiment shown on the left, and 250CD8� cells per well were cultured in that shown on the right. Spots were counted in triplicate in each of the 8 experimental groups, and the barsrepresent mean values and SEs. The unpaired Student’s t test was used for statistical analysis. *P � .01 and **P � .02. B, Detection of human CD8�
cells that produce IFN-� in response to stimulation with an EBV-positive LCL by flow cytometry. CD8� cells were isolated from the spleen of anEBV-infected mouse and cocultured with the autologous LCL. Intracellular IFN-� was stained and analyzed as described in Methods.
New Mouse Model of EBV Infection ● JID 2008:198 (1 September) ● 679
man T cells). This increase in CD8� cells were even more con-
spicuous when their definite number was counted (figure 3B).
When hNOG mice were inoculated with serially diluted virus
samples, a striking dose response was evident; mice inoculated
with higher doses exhibited a more profound increase in CD8�
cells at earlier time points (figure 3C). Further flow cytometry
analyses showed that CD45RO� memory T cells, compared with
CD45RA� T cells, increased in infected hNOG mice (figure 3D).
Expression of a T cell activation marker, HLA-DR, was observed
mainly in CD8� cells rather than in CD4� cells (figure 3D).
To demonstrate that these CD8� T cells were directed against
EBV-infected cells, we examined IFN-� secretion after stimula-
tion with EBV-transformed cells. For this purpose, we first es-
tablished an LCL using B cells isolated from the same cord blood
that was used to isolate HSCs for transplantation. CD8� T cells,
isolated from the peripheral blood of EBV-infected hNOG mice,
were incubated with this autologous LCL, and cells secreting
IFN-� were detected by ELISPOT assay. For all 3 EBV-infected
hNOG mice (which had been infected at 1 � 103 TD50) thus
examined, a significant number of spots were recognized in the
wells in which CD8� T cells were mixed with the autologous
LCL, whereas those cells incubated with unrelated LCL had
many fewer spots (data from 2 mice are shown in figure 4A).
CD8� T cells isolated from uninfected hNOG mice did not give
a significant number of spots (data not shown). Release of IFN-�
was blocked by antibody specific to human major histocompat-
ibility complex (MHC) class I but not by that specific to human
MHC class II (figure 4A). These results clearly show that a T cell
response restricted by human MHC class I was mounted against
EBV-infected cells. In addition, in 5 of the 6 EBV-infected
hNOG mice examined (infected at 1 � 103 TD50), flow cytom-
etry also demonstrated production of IFN-� by CD8� T cells
isolated from the spleen and stimulated with an autologous LCL
(figure 4B).
EBV-specific antibody response in hNOG mice. Serum
samples from 30 EBV-infected hNOG mice were examined by
Western blotting for IgM antibodies reactive with a bacterially
expressed GST-BFRF3 fusion protein. The BFRF3 protein is a
major component of the virus capsid antigen of EBV [26]. The
results are shown in figure 5A and indicated that four serum
samples (from mice infected at 1 � 101 or 1 � 103 TD50) con-
tained IgM antibody reactive with it. These serum samples re-
acted also with the 18-kDa BFRF3-encoded protein in the lysate
of Akata cells stimulated with IgG antibody to activate virus pro-
duction (figure 5B). Similar experiments with human IgG–spe-
cific secondary antibody did not show a positive reaction with
either GST-BFRF or p18BFRF3. Six serum samples collected from
uninfected hNOG mice reacted with neither the 18-kDa protein
nor GST-BFRF3 (figure 5 and data not shown). These results
indicate that hNOG mice have the ability to mount an IgM re-
sponse to EBV.
DISCUSSION
The lymphoproliferative disease induced in hNOG mice is re-
markably similar to the human lymphoproliferative disorder
seen in immunocompromised hosts [27] with respect to histol-
ogy, surface phenotype, and the type of EBV gene expression
Figure 5. Demonstration of IgM antibody to the Epstein-Barr virus (EBV) BFRF3 protein in the serum of humanized NOG (hNOG) mice. A, Immunoblotwith the glutathione s-transferase (GST)–BFRF3 fusion protein. Purified GST-BFRF3 fusion protein was examined with serum from an EBV-uninfectedperson (a), an EBV-infected person (b), EBV-infected hNOG mice (c and d ), and an uninfected hNOG mice (e and f ). B, Immunoblot with the lysate ofEBV-producing Akata cells. Lysate of anti-IgG–treated Akata cells, labeled Akata(�), and of EBV-negative Akata cells, labeled Akata(�), was examinedusing serum from an EBV-uninfected person (a), an EBV-infected person (b), EBV-infected hNOG mice (c and d ), and uninfected hNOG mice (e and f ).
680 ● JID 2008:198 (1 September) ● Yajima et al.
(latency III). Reproduction of latency III in the present study
makes for an interesting contrast with the previous model using
NOD/scid mice, which exhibited the latency II pattern [10].
EBV infection in lower doses resulted in a transient increase in
EBV DNA load in the peripheral blood, followed by apparently
asymptomatic infection that persisted for at least 22 weeks. This
type of asymptomatic EBV infection has not been described in
nonprimate models of EBV infection and may be regarded as a
model of human EBV latency. To compare this condition in
NOG mice with EBV latency in humans precisely, we need to
further investigate the nature of host cells (i.e., whether they are
memory B cells), the pattern of EBV gene expression in them,
and the involvement of anti-EBV immune responses in its main-
tenance.
In hNOG mice, human T cells develop in thymus tissue, in
which epithelial cells are of murine origin [16]. It is therefore
interesting that they could mount a T cell response restricted by
human MHC class I. Although this suggests that positive selec-
tion of human T cells occurred in hNOG mice, the mechanism of
T cell education remains unclear. Alloantigen-specific and hu-
man MHC class I–restricted T cell cytotoxicity has been reported
in hNOG mice [15, 16]. An EBV-induced T cell response was
evident in mice that received high doses of virus and developed
lymphoproliferative disorder, suggesting that the T cell response
in hNOG mice was not sufficient to control EBV-induced lym-
phoproliferation when they were infected at high doses. That
only a minor fraction of CD8� T cells appeared to be EBV spe-
cific, as evidenced by ELISPOT assay and flow cytometry, may
explain this result, at least partially. A humoral immune re-
sponse to EBV has not been documented in previous mouse
models of EBV infection, and therefore the NOG mouse may
provide a valuable tool to analyze the mechanism and the pro-
tective roles of antibody response in EBV infection. We have to
date clearly identified only IgM antibody to the 18-kDa compo-
nent of virus capsid antigen in a minor fraction (4/30) of infected
mice. We are currently attempting to improve sensitivity and to
see whether hNOG mice can mount a more efficient and diver-
gent antibody response to the virus, possibly including the pro-
duction of IgG antibodies. Because both the T cell–mediated and
the humoral immune response are elicited in hNOG mice, they
may be useful in the evaluation of candidate EBV vaccines.
Very recently, humanized mice based on other immunodefi-
cient mouse strains were prepared, and EBV was used as a typical
pathogen to analyze their immune functions. Traggiai et al. [12]
infected humanized Rag2�/�IL-2R��/� mice with EBV and doc-
umented an in vitro proliferative response by CD8� T cells to an
autologous LCL. Melkus et al. [11], on the other hand, human-
ized NOD/scid mice by transplanting human fetal liver, thymus,
and HSCs and succeeded in inducing an EBV-specific T cell im-
mune response as well as an innate immune response to toxic
shock syndrome toxin 1. These 2 studies were performed mainly
using immunological standpoints and did not provide detailed
data from virological investigations. An advantage of the NOG
mouse model described here is that it does not require a fine
surgical procedure using human fetal tissue; therefore, NOG
mice can be easily provided in large quantities.
In immunocompromised humans, failure of immunosurveil-
lance may lead to the development of lymphoproliferative dis-
order. We expect that the NOG mouse model can be used to
analyze the exact relationship between immunodeficiency and
the development of lymphoproliferative disorder. Immune re-
sponses in the hNOG mouse can be modulated by immunosup-
pressive drugs (such as cyclosporine A) or HIV, and the devel-
opment of lymphoproliferative disorder can be analyzed with
special reference to the nature and level of immunodeficiency.
This kind of study, which has not been possible with conven-
tional scid mice, may reveal an exact condition in which lympho-
proliferative disorder develops and may thereby aid the develop-
ment of a specified immunosuppressive procedure that evades
this condition and precludes the risk of lymphoproliferative dis-
order.
In summary, the NOG mouse is able to recapitulate various
essential elements of human EBV infection and is therefore, to
our knowledge, the most comprehensive small-animal model of
EBV infection described to date. It should be a valuable tool for
the study of the pathogenesis, prevention, and treatment of EBV
infection.
Acknowledgments
We thank Satoshi Itakura, Fuyuko Kawano, Eri Yamada, Miki Mizukami,and Ken Watanabe for technical assistance. We thank Shizuko Minegishi foradvice on flow cytometry, Atsushi Komano for advise on the enzyme-linkedimmunospot assay, Ayako Demachi-Okamura and Kiyotaka Kuzushima foradvice on detection of Epstein-Barr virus–specific T cells, and Shosuke Imaifor helpful discussions. We thank the Tokyo Cord Blood Bank for supplyingcord blood.
References
1. Rickinson AB, Kieff E. Epstein-Barr virus. In: Knipe DM, Howley PM,eds. Fields virology. Philadelphia: Lippincott Williams & Wilkins, 2001:2575– 628.
2. Kieff E, Rickinson AB. Epstein-Barr virus and its replication. In: KnipeDM, Howley PM, eds. Fields virology. 4th ed. Philadelphia: LippincotWilliams & Wilkins, 2001:2511–74.
3. Young LS, Finerty S, Brooks L, Scullion F, Rickinson AB, Morgan AJ.Epstein-Barr virus gene expression in malignant lymphomas induced byexperimental virus infection of cottontop tamarins. J Virol 1989; 63:1967–74.
4. Miller G, Shope T, Coope D, et al. Lymphoma in cotton-top marmosetsafter inoculation with Epstein-Barr virus: tumor incidence, histologicspectrum antibody responses, demonstration of viral DNA, and charac-terization of viruses. J Exp Med 1977; 145:948 – 67.
5. Cho Y, Ramer J, Rivailler P, et al. An Epstein-Barr-related herpesvirusfrom marmoset lymphomas. Proc Natl Acad Sci USA 2001; 98:1224 –9.
6. Moghaddam A, Rosenzweig M, Lee-Parritz D, Annis B, Johnson RP,Wang F. An animal model for acute and persistent Epstein-Barr virusinfection. Science 1997; 276:2030 –3.
7. Mosier DE, Gulizia RJ, Baird SM, Wilson DB. Transfer of a functionalhuman immune system to mice with severe combined immunodefi-ciency. Nature 1988; 335:256 –9.
New Mouse Model of EBV Infection ● JID 2008:198 (1 September) ● 681
8. Okano M, Taguchi Y, Nakamine H, et al. Characterization of Epstein-Barr virus-induced lymphoproliferation derived from human periph-eral blood mononuclear cells transferred to severe combined immuno-deficient mice. Am J Pathol 1990; 137:517–22.
9. Rowe M, Young LS, Crocker J, Stokes H, Henderson S, Rickinson AB.Epstein-Barr virus (EBV)-associated lymphoproliferative disease in theSCID mouse model: implications for the pathogenesis of EBV-positivelymphomas in man. J Exp Med 1991; 173:147–58.
10. Islas-Ohlmayer M, Padgett-Thomas A, Domiati-Saad R, et al. Experi-mental infection of NOD/SCID mice reconstituted with human CD34�cells with Epstein-Barr virus. J Virol 2004; 78:13891–900.
11. Melkus MW, Estes JD, Padgett-Thomas A, et al. Humanized micemount specific adaptive and innate immune responses to EBV andTSST-1. Nat Med 2006; 12:1316 –22.
12. Traggiai E, Chicha L, Mazzucchelli L, et al. Development of a humanadaptive immune system in cord blood cell-transplanted mice. Science2004; 304:104 –7.
13. Hiramatsu H, Nishikomori R, Heike T, et al. Complete reconstitution ofhuman lymphocytes from cord blood CD34� cells using the NOD/SCID/�c
null mice model. Blood 2003; 102:873– 80.14. Ito M, Hiramatsu H, Kobayashi K, et al. NOD/SCID/�c
null mouse: anexcellent recipient mouse model for engraftment of human cells. Blood2002; 100:3175– 82.
15. Yahata T, Ando K, Nakamura Y, et al. Functional human T lymphocytedevelopment from cord blood CD34� cells in nonobese diabetic/Shi-scid, IL-2 receptor gamma null mice. J Immunol 2002; 169:204 –9.
16. Ishikawa F, Yasukawa M, Lyons B, et al. Development of functionalhuman blood and immune systems in NOD/SCID/IL2 receptor �chainnull mice. Blood 2005; 106:1565–73.
17. Miyazato P, Yasunaga J, Taniguchi Y, Koyanagi Y, Mitsuya H, MatsuokaM. De novo human T-cell leukemia virus type 1 infection of humanlymphocytes in NOD-SCID, common gamma-chain knockout mice.J Virol 2006; 80:10683–91.
18. Watanabe S, Terashima K, Ohta S, et al. Hematopoietic stem cell-engrafted NOD/SCID/IL2Rgamma null mice develop human lymphoidsystems and induce long-lasting HIV-1 infection with specific humoralimmune responses. Blood 2007; 109:212– 8.
19. Dewan MZ, Terashima K, Taruishi M, et al. Rapid tumor formation ofhuman T-cell leukemia virus type 1-infected cell lines in novel NOD-SCID/gammacnull mice: suppression by an inhibitor against NF-kappaB.J Virol 2003; 77:5286 –94.
20. Watanabe S, Ohta S, Yajima M, et al. Humanized NOD/SCID/IL2R�null
mice transplanted with hematopoietic stem cells under nonmyeloabla-tive conditions show prolonged life spans and allow detailed analysis ofhuman immunodeficiency virus type 1 pathogenesis. J Virol 2007; 81:13259 – 64.
21. Takada K, Ono Y. Synchronous and sequential activation of latentlyinfected Epstein-Barr virus genomes. J Virol 1989; 63:445–9.
22. Condit RC. Principles of virology. In: Knipe DM, Howley PM, eds.Fields virology. Philadelphia: Lippincott Williams & Wilkins, 2001:19 –51.
23. Kimura H, Morita M, Yabuta Y, et al. Quantitative analysis of Epstein-Barr virus load by using a real-time PCR assay. J Clin Microbiol 1999;37:132– 6.
24. Nakamura H, Iwakiri D, Ono Y, Fujiwara S. Epstein-Barr-virus-infectedhuman T-cell line with a unique pattern of viral-gene expression. Int JCancer 1998; 76:587–94.
25. Kuzushima K, Hoshino Y, Fujii K, et al. Rapid determination of Epstein-Barr virus-specific CD8� T-cell frequencies by flow cytometry. Blood1999; 94:3094 –100.
26. van Grunsven WM, Nabbe A, Middeldorp JM. Identification and mo-lecular characterization of two diagnostically relevant marker proteinsof the Epstein-Barr virus capsid antigen complex. J Med Virol 1993; 40:161–9.
27. Rezk SA, Weiss LM. Epstein-Barr virus-associated lymphoproliferativedisorders. Hum Pathol 2007; 38:1293–304.
682 ● JID 2008:198 (1 September) ● Yajima et al.